{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://gateway.jrich.ai/schemas/wakeword-training-request-v1.json",
  "title": "Resolved Wakeword Training Request v1",
  "type": "object",
  "additionalProperties": false,
  "required": [
    "schema_version",
    "primary_phrase",
    "variants",
    "near_misses",
    "evaluation_profile",
    "source_lock_path",
    "source_lock_sha256",
    "intended_release_id",
    "runner_image_digest",
    "pipeline_config",
    "run_kind",
    "derived_from_run_id"
  ],
  "properties": {
    "schema_version": {"const": 1},
    "primary_phrase": {"type": "string", "minLength": 1},
    "variants": {"type": "array", "items": {"type": "string", "minLength": 1}},
    "near_misses": {"type": "array", "items": {"type": "string", "minLength": 1}},
    "evaluation_profile": {"type": "string", "minLength": 1},
    "source_lock_path": {"type": "string", "minLength": 1},
    "source_lock_sha256": {"type": "string", "pattern": "^[a-f0-9]{64}$"},
    "intended_release_id": {"type": "string", "minLength": 1},
    "runner_image_digest": {"type": "string", "pattern": "^sha256:[a-f0-9]{64}$"},
    "pipeline_config": {
      "type": "object",
      "additionalProperties": false,
      "required": ["model_name", "target_phrases"],
      "properties": {
        "model_name": {"type": "string", "pattern": "^[A-Za-z0-9][A-Za-z0-9_-]*$"},
        "family_slug": {"type": "string", "pattern": "^[a-z0-9]+(?:-[a-z0-9]+)*$"},
        "version": {"type": "integer", "minimum": 1},
        "threshold": {
          "oneOf": [{"type": "number", "minimum": 0, "maximum": 1}, {"const": "auto"}]
        },
        "target_phrases": {"type": "array", "minItems": 1, "items": {"type": "string", "minLength": 1}},
        "custom_negative_phrases": {"type": "array", "items": {"type": "string", "minLength": 1}},
        "n_samples": {"type": "integer", "minimum": 1},
        "n_samples_val": {"type": "integer", "minimum": 1},
        "n_background_samples": {"type": "integer", "minimum": 1},
        "n_background_samples_val": {"type": "integer", "minimum": 1},
        "tts_batch_size": {"type": "integer", "minimum": 1},
        "max_speakers": {"type": ["integer", "null"], "minimum": 1},
        "tts_backend": {"const": "piper_vits"},
        "noise_scales": {"type": "array", "minItems": 1, "items": {"type": "number"}},
        "noise_scale_ws": {"type": "array", "minItems": 1, "items": {"type": "number"}},
        "length_scales": {"type": "array", "minItems": 1, "items": {"type": "number"}},
        "slerp_weights": {"type": "array", "minItems": 1, "items": {"type": "number"}},
        "augmentation": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "clip_duration": {"type": "number", "exclusiveMinimum": 0},
            "batch_size": {"type": "integer", "minimum": 1},
            "rounds": {"type": "integer", "minimum": 1},
            "background_paths": {
              "type": "array",
              "items": {
                "type": "string",
                "pattern": "^/inputs/(?!\\.\\.(?:/|$))(?!.*?/\\.\\.(?:/|$)).+"
              }
            },
            "rir_paths": {
              "type": "array",
              "items": {
                "type": "string",
                "pattern": "^/inputs/(?!\\.\\.(?:/|$))(?!.*?/\\.\\.(?:/|$)).+"
              }
            }
          }
        },
        "model": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "model_type": {"enum": ["dnn", "rnn", "conv_attention"]},
            "model_size": {"enum": ["tiny", "small", "medium", "large"]}
          }
        },
        "steps": {"type": "integer", "minimum": 1},
        "learning_rate": {"type": "number", "minimum": 0},
        "weight_decay": {"type": "number", "minimum": 0},
        "label_smoothing": {"type": "number", "minimum": 0},
        "max_negative_weight": {"type": "number", "minimum": 0},
        "target_fp_per_hour": {"type": "number", "minimum": 0},
        "batch_n_per_class": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "positive": {"type": "integer", "minimum": 1},
            "adversarial_negative": {"type": "integer", "minimum": 1},
            "ACAV100M_sample": {"type": "integer", "minimum": 1},
            "background_noise": {"type": "integer", "minimum": 1}
          }
        }
      }
    },
    "run_kind": {"enum": ["commercial", "local_experiment", "quarantined_smoke"]},
    "derived_from_run_id": {
      "type": ["string", "null"],
      "pattern": "^[a-f0-9]{32}$"
    }
  }
}